Bounded scrapes
Up to 10 of the top-ranking pages are fetched once per brief, then discarded after the statistics are computed.
Content briefs
Write against what actually ranked on the day.
A brief pulls the stored top-ten result set, scrapes a bounded number of those pages, computes corpus statistics, and drafts an outline where every item cites the evidence it came from.
Evidence before outline
Statistics first, then a draft that cites them.
Nothing enters the outline that cannot point at a scraped document, a statistic, a stored question, or a cluster term.
Deterministic corpus statistics
Word-count range and average, a heading histogram, and an inventory of the structured-data types those pages declared.
An outline that cites
Every outline node and question points at a stored document, statistic, People Also Ask row, or cluster term, or it is dropped.
Draft editor with re-scoring
Write in the editor and re-score against the same statistics, so the target you are measured on never moves under you.
Product truth
A snapshot of pages scraped on a stated date.
This illustrative frame mirrors normalized output. It is not a live search.
| Signal | What the product shows | Source |
|---|---|---|
| Corpus word count | Range and average across the pages scraped for this brief | Provider observation |
| Outline node | Carries the citation that justified it | AI interpretation |
| Draft score | Measured against this brief's own statistics | AI interpretation |
How it runs
One unit, one rolling ceiling, one snapshot.
Cadence, source, and spend ceilings stay visible beside every brief.
On demand
Briefs run when you ask. Reopening a brief or its draft history is a read, never a second set of scrapes.
Top 10 results plus scrapes
The result set comes from storage when it is fresh enough, and the page text from a bounded scrape of those same URLs.
A rolling cost ceiling
A run halts if its accumulated vendor and AI cost would exceed the per-run ceiling, and tells you what it completed before stopping.
What we don't doHonest limit Honest limit Honest limit
Three things this feature will never claim.
Every limit below is enforced in the product, not written on a disclaimer page.
A dated snapshot, not a timeless truth
Statistics describe the pages scraped on a stated date — never a timeless "what the top ten look like" claim. A run costs one brief whether the result set came from storage or a fresh check.
Cite or abstain
Every outline node and question must cite supplied evidence or it is dropped, and a brief with nothing to cite abstains. The product never invents a source to fill an outline.
Guidance, and nothing is published
A brief is guidance, not a likelihood or guarantee of ranking, and nothing is published to your site. On Pro, briefs are reachable only through their credit pack.
Every limit below is enforced by the product, not a disclaimer.
Evidence-backed drafts
Brief the writer with what ranked.
Create an account to build your first brief, or begin with a free site audit.